A Review of Dengue & Malaria Detection Systems from PBS using Machine Learning

Kunal Vinay Gohad , Dr Rajshree Pankaj Wankhade

Digital Image Processing, Machine Learning, SVM, CNN, Disease Detection, Biomedical, Dengue, Malaria.

Mosquito-borne diseases such as Dengue and Malaria remain major global public health challenges, causing millions of infections and substantial mortality each year, particularly in tropical and subtropical regions. Conventional diagnostic methods for both diseases rely heavily on manual microscopic examination of peripheral blood smears, a process that is labor-intensive, time-consuming, and vulnerable to observer variability. Recent advances in digital pathology and image-processing-based automation offer promising solutions for early, accurate, and cost-effective diagnosis. In dengue detection, peripheral blood smear (PBS) analysis reveals hallmark changes such as thrombocytopenia and lymphocyte morphological alterations. Machine-learning approaches using morphological and Gray-Level Spatial Dependence Matrix (GLSDM) features extracted from lymphocyte nuclei have demonstrated high diagnostic accuracy, with Support Vector Machine-based models achieving up to 95.74% accuracy in distinguishing dengue-infected samples. Similarly, for malaria diagnosis, automated image analysis systems employing color thresholding, segmentation, blob analysis, watershed techniques, and shape-based classification have shown excellent performance in detecting Plasmodium-infected red blood cells. Systems using texture and intensity features, neural network classifiers, and morphological processing have achieved accuracies ranging from 93% to 99%, significantly reducing dependency on expert microscopists. This combined study highlights the growing potential of AI-driven digital microscopy for the automated screening of vector-borne diseases. Integrating machine learning, deep learning, and advanced image-processing techniques enables rapid detection of infected cells, accurate parasite quantification, and improved clinical decision-making. Such automated frameworks can provide scalable, low-cost diagnostic support in resource-limited settings, thereby strengthening global disease surveillance and early intervention strategies for both dengue and malaria.

Volume 14 Issue 2, February-2026

Pages : e749-e759